Fixed the call to CalcNormFloat to use neweve[]->co.

Goofster: The difference between you and Ton reporting the problem was
that he actually took the time to analyze the warning and pin down the
problem, while your "hahaha, your code has a warning!" didn't really
help me a lot in the first place.

source/blender/src/editmesh.c

@@ -8392,7 +8392,7 @@ void bevel_mesh(float bsize, int allfaces)
 				}
 				if(evl) {
 					float inp;	
-					CalcNormFloat(neweve[0], neweve[1], neweve[2], evl->n);
+					CalcNormFloat(neweve[0]->co, neweve[1]->co, neweve[2]->co, evl->n);
 					inp= evl->n[0]*G.vd->viewmat[0][2] + evl->n[1]*G.vd->viewmat[1][2] + evl->n[2]*G.vd->viewmat[2][2];
 					if(inp < 0.0) flipvlak(evl);
 #ifdef BEV_DEBUG